Influence of a partially oxidized calcium cathode on the performance ofpolymeric light emitting diodesG. G. Andersson, M. ...
JOURNAL OF APPLIED PHYSICS                                                         VOLUME 90, NUMBER 3                    ...
J. Appl. Phys., Vol. 90, No. 3, 1 August 2001                                                                             ...
1378      J. Appl. Phys., Vol. 90, No. 3, 1 August 2001                                                                   ...
J. Appl. Phys., Vol. 90, No. 3, 1 August 2001                                                                             ...
1380      J. Appl. Phys., Vol. 90, No. 3, 1 August 2001                                                                   ...
J. Appl. Phys., Vol. 90, No. 3, 1 August 2001                                                                             ...
1382     J. Appl. Phys., Vol. 90, No. 3, 1 August 2001                                                                    ...
Upcoming SlideShare
Loading in …5

Influence of a_partially_oxidized_calcium_cathode_on_the_performance_of_polymeric_light_emitting_diodes


Published on

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Influence of a_partially_oxidized_calcium_cathode_on_the_performance_of_polymeric_light_emitting_diodes

  1. 1. Influence of a partially oxidized calcium cathode on the performance ofpolymeric light emitting diodesG. G. Andersson, M. P. de Jong, F. J. J. Janssen, J. M. Sturm, L. J. van IJzendoorn et al.Citation: J. Appl. Phys. 90, 1376 (2001); doi: 10.1063/1.1383577View online: Table of Contents: by the American Institute of Physics.Related ArticlesInfluence of laser lift-off on optical and structural properties of InGaN/GaN vertical blue light emitting diodesAIP Advances 2, 022122 (2012)Identification of device degradation positions in multi-layered phosphorescent organic light emitting devices usingwater probesAppl. Phys. Lett. 100, 183306 (2012)Efficiency enhancement of blue InGaN/GaN light-emitting diodes with an AlGaN-GaN-AlGaN electron blockinglayerJ. Appl. Phys. 111, 094503 (2012)Identification of device degradation positions in multi-layered phosphorescent organic light emitting devices usingwater probesAPL: Org. Electron. Photonics 5, 102 (2012)High efficiency warm-white organic light emitting diodes from a single emitter in graded-doping devicearchitectureAPL: Org. Electron. Photonics 5, 101 (2012)Additional information on J. Appl. Phys.Journal Homepage: Information: downloads: for Authors: Downloaded 08 May 2012 to Redistribution subject to AIP license or copyright; see
  2. 2. JOURNAL OF APPLIED PHYSICS VOLUME 90, NUMBER 3 1 AUGUST 2001Influence of a partially oxidized calcium cathode on the performanceof polymeric light emitting diodes G. G. Andersson, M. P. de Jong, F. J. J. Janssen, J. M. Sturm, L. J. van IJzendoorn,a) A. W. Denier van der Gon, M. J. A. de Voigt, and H. H. Brongersma Faculty of Applied Physics, Eindhoven University of Technology, Postbox 513, 5600 MB Eindhoven, The Netherlands ͑Received 23 October 2000; accepted for publication 8 May 2001͒ We investigated the influence of the presence of oxygen during the deposition of the calcium cathode on the structure and on the performance of polymeric light emitting diodes ͑pLEDs͒. The oxygen background pressure during deposition of the calcium cathode of polymeric LEDs was varied. Subsequently, the oxygen depth distribution was measured and correlated with the performance of the pLEDs. The devices have been fabricated in a recently built ultraclean setup. The polymer layers of the pLEDs have been spincoated in a dry nitrogen atmosphere and transported directly into an ultrahigh vacuum chamber where the metal electrodes have been deposited by evaporation. We used indium–tin–oxide as anode, OC1C10 PPV as electroluminescent polymer, calcium as cathode, and aluminum as protecting layer. We achieved reproducibility of about 15% in current and brightness for devices fabricated in an oxygen atmosphere of Ӷ10Ϫ9 mbar. For further investigations the calcium deposition was carried out in an oxygen atmosphere from 10Ϫ8 to 10Ϫ5 mbar. We determined the amount of oxygen in the different layers of the current–voltage-light characterized pLEDs with elastic recoil detection analysis and correlated it with the characteristics of the devices. The external efficiency of the pLEDs decreases continuously with increasing oxygen pressure, the current shows a pronounced minimum. The brightness mostly decreases with increasing oxygen with an indication of a slight minimum. PLEDs with completely oxidized calcium are not operational. The first contact of the pLEDs with the dry glove box environment leads to an immediate reduction of current and brightness which is caused by the cooling of the devices by several degrees. Determining reproducible characteristics of pLEDs in the vacuum requires the measurement of their temperature. © 2001 American Institute of Physics. ͓DOI: 10.1063/1.1383577͔I. INTRODUCTION polymeric LEDs with calcium as cathode deposited under high vacuum conditions might be attributed to the presence Polymeric light-emitting diodes ͑pLEDs͒ are fabricated of oxygen. Nevertheless, using magnesium as an electrodeunder a broad range of conditions. The influence of the fab- they achieved the best LEDs at the lowest oxygen pressure,rication conditions on the characteristics of the pLEDs has i.e., at 10Ϫ10 mbar. 6 They also described the degradation ofbeen partially investigated, but full understanding has not the structure of PPV due to photo-oxidation.7 Bohler et al. ¨been reached yet. In the fabrication process of pLEDs the found an increase of current, brightness, and efficiency inlow work function cathode is generally deposited by evapo- LEDs with a tris͑8-hydroxyquinolinato͒ ͑Alq3͒ electrolumi-ration in a high vacuum system.1– 4 Unintentional impurities nescent layer and magnesium cathode if the pressure duringsuch as oxygen and moisture are present in a broad range of deposition of the cathode decreases to ultrahigh vacuumdifferent concentrations. Commonly used materials for the ͑UHV͒ conditions.8 Summarizing, the influence on the pres-cathodes are calcium or magnesium. The presence of oxygen ence of oxygen and moisture and the influence of theand moisture will influence the work function of the cathode vacuum conditions during the deposition of magnesium andand thus the balance of the charge carrier or the chemical calcium cathodes has been partially investigated, but the cor-interaction of the low work function metal with the electrolu- relation with the resulting modification of the structure of the ¨minescent polymer. Broms et al. found that an oxygen pres- LEDs and the reasons for the influence of the performance Ϫ6sure of about 10 mbar during evaporation of the calcium have not been found. However, it is of major interest for thecathode is necessary to achieve operational pLEDs using fabrication process to know which vacuum conditions areindium–tin–oxide ͑ITO͒ and calcium as electrodes and cy- necessary to archive the best performance of LEDs.anosubstituted polyphenylenevinylene ͑CNPPV͒ as lumines- Recently, we built a setup for the fabrication of pLEDscent polymer,5 while deposition of the calcium at lower pres- under controlled conditions. Oxygen and moisture can besures lead to device failure. Thus, the good performance of kept at a very low level during the fabrication and investiga- tion of the LEDs. We investigated the influence of the pres-a͒ Author to whom correspondence should be addressed; electronic mail: ence of oxygen during the deposition of the calcium cathode on the structure and on the device characteristic of polymeric0021-8979/2001/90(3)/1376/7/$18.00 1376 © 2001 American Institute of Physics Downloaded 08 May 2012 to Redistribution subject to AIP license or copyright; see
  3. 3. J. Appl. Phys., Vol. 90, No. 3, 1 August 2001 Andersson et al. 1377 FIG. 1. Scheme of the LED production facility. FIG. 2. Scheme of the investigated structure.LEDs. Oxygen is expected to have a strong influence on theperformance of LEDs using reactive low work function met-als such as magnesium and calcium. LEDs is determined with OPT301 photodiodes from Burr The influence of the presence of oxygen during the Brown which have been calibrated with a luminescencedeposition of the calcium cathode was investigated by meter LS 110 from Minolta. The temperature of the samplesevaporating the calcium in an oxygen atmosphere of during operation can be measured with NTCs and with a10Ϫ8 – 10Ϫ5 mbar. We determined both the current–voltage LUXTRON Fluoroptic thermometer. The temperature of the(I – V)-light characteristics and the concentration of oxygen. devices in the glove box is 23Ϯ2 °C.The concentration depth profiles have been determined with The LEDs or the half fabricates can be transported underelastic recoil detection analysis ͑ERDA͒. We found a strong dry nitrogen atmosphere to setups where the elemental com-correlation between the presence of oxygen in the calcium position of the surface and the bulk can be analyzed. Forcathode and the characteristics of the devices. The best LEDs analysis of the surface we apply low energy ion scatteringwere obtained with the lowest possible concentration of oxy- and x-ray photoelectron spectroscopy. For analysis of thegen present during calcium evaporation. bulk RBS, ERDA, and particle induced x-ray emission are applied.II. EXPERIMENT We used ITO as the anode, a 130 nm thick layer of the electroluminescent polymer OC1C10-PPV as the emissiveA. pLED fabrication and IV-light characterization layer, 80 nm calcium as the cathode, and aluminum as the The devices are fabricated in a closed setup sketched in protecting layer. The glass/ITO substrates ͑100 nm ITO, 30Fig. 1. Six devices can be fabricated at the same time under ⍀/ᮀ, Merck͒ were cleaned in ultrapure acetone ͑Uvasolidentical conditions. The setup consists of a glove box with from Merck͒ and 2-propanol ͑Secsolv from Merck͒ each forless than 1 ppm oxygen and moisture, an UV-ozone pretreat- 10 min in an ultrasonic bath. This wet cleaning step wasment chamber and an UHV evaporation chamber. The setup followed by an UV-ozone treatment for 20 min. Directly af-is constructed such that samples may be transferred between ter this treatment, the samples were transferred to the glovethe different chambers without exposure of the samples to box with a dry nitrogen atmosphere without being exposed toair. The UV-ozone treatment is useful to remove hydrocar- air. The OC1C10-PPV layer was spin coated in the glove boxbons from the surface of ITO substrates.9 The polymer layers from a 0.71 wt % toluene solution. The samples were trans-are spin coated in the glove box, which is equipped with an ferred into an UHV evaporation chamber without havingactivated carbon filter in order to remove the evaporated or- contact to air. The calcium cathode and aluminum protectionganic solvents. The window of the glove box is covered with layers were evaporated from effusion cells at temperatures ofa foil in order to prevent the polymers from the exposure to 510 and 1150 °C, respectively, the deposition rates at theselight of short wavelengths. The samples are introduced via a temperatures are for calcium 4 Å/s and aluminum 1 Å/s. At atransfer chamber ͑base pressure of 5ϫ10Ϫ8 mbar͒ into the partial pressure of Ͻ10Ϫ9 mbar oxygen the concentration ofUHV ͑base pressure of 1ϫ10Ϫ9 mbar͒ chamber. The partial oxygen will be less than 0.1 at. % in the calcium layer at thispressures in the UHV are measured with a residual gas ana- deposition rate. Oxygen can be introduced into the UHVlyzer Leybold Quadruvac Q 100 with a detection limit of chamber by means of a leakage valve. The pressure duringϳ10Ϫ9 mbar. The UHV chamber is equipped with an infra- deposition of calcium was varied from less than 10Ϫ9 tored lamp for soft bake of the organic layers. We can evapo- 10Ϫ5 mbar. The fabrication of the devices was finished byrate three different metals from commercial Riber ABN 135 evaporation of the aluminum layer. A scheme of the devicesL effusion cells. The layer thicknesses are determined with is shown in Fig. 2. 1 h after finishing the evaporation of thean Intellemetrics IL 150 quartz crystal monitor, which was aluminum layer the devices have a temperature of 45Ϯ2 °C.calibrated with Rutherford backscattering spectrometry Within 15 h the equilibrium temperature in the UHV of͑RBS͒. The temperatures of the samples are measured during about 28Ϯ2 °C is reached.evaporation with NTC thermistors. IV-light characteristics of From each set of LEDs we investigated oxygen and car-the LEDs can be measured either directly after fabrication in bon concentration depth profiles with ERDA at cryogenicthe UHV chamber or in the glove box. The brightness of the temperatures. Downloaded 08 May 2012 to Redistribution subject to AIP license or copyright; see
  4. 4. 1378 J. Appl. Phys., Vol. 90, No. 3, 1 August 2001 Andersson et al.FIG. 3. I – V-light characteristics of LEDs with deposition of the calcium FIG. 4. Current and brightness for different partial pressures of oxygen at acathode in an oxygen pressure of Ͻ10Ϫ9 and 5ϫ10Ϫ7 mbar. The charac- bias of 5 V. 10Ϫ9 mbar means that no oxygen was let into the UHV chamberteristics are measured in the glove box. during deposition of the calcium, i.e., the oxygen pressure was some orders of magnitude lower than 10Ϫ9 mbar. At an oxygen pressure around 5 ϫ10Ϫ6 mbar the current and brightness of the devices changes rapidly with the pressure. Due to their position with respect to the oxygen inlet theB. Oxygen depth profiling with cryogenic ERDA samples marked with # experienced a somewhat higher oxygen pressure, and we attribute the difference between the groups thus to the high sensi- ERDA is a suitable method to determine concentration tivity of the device performance on the oxygen pressure in this range. Dif-depth profiles of light elements such as carbon and oxygen ferences greater than 15% within one set of devices were observed only inup to a depth of approximately 1 ␮m. A depth resolution of this case, for all other experiments it was less than that.10 nm can be achieved. The pLEDs were transferred in drynitrogen to our cryogenic RBS/ERDA setup.10 This setup hasbeen especially developed for RBS/ERDA analysis of or- For devices that are prepared under comparable conditions,ganic samples, which are in general very sensitive to ion these current and brightness values are reproducible with anirradiation. Ion irradiation of organic materials leads to the accuracy of 15% at bias voltages у4 V.formation of small volatile species along the ion track. Cool- The characteristics of devices that were fabricated in aning the samples to cryogenic temperatures reduces drastically oxygen pressure у10Ϫ8 mbar show a decrease of the bright-the mobility of these species. Measuring at room temperature ness with increasing oxygen pressure, but the current is re-would distort the concentration depth profiles, whereas the duced only in a certain pressure regime. Current and bright-influence of the ion irradiation at cryogenic temperatures is ness of the devices at a bias of 5 V are plotted in Fig. 4. Thenegligible. In our cryogenic setup, the samples are cooled data points plotted at the x value of 10Ϫ9 mbar correspond towith a Gifford–McMahon cryocooler by APD cryogenics LEDs that were prepared without admitting oxygen to theInc, which has a cooling power of 2 W at 10 K. The sample UHV chamber during deposition of the calcium ͑10Ϫ9 mbartemperature depends critically on the heat load transferred by is the detection limit for oxygen for the residual gas ana-the ion beam and the thermal contact between the samples lyzer͒. The oxygen pressure in an UHV system with the men-and the cooler. For the experiments reported in this article tioned base pressure is usually about several orders of mag-the sample temperature is estimated as less than 30 K. At nitude lower than 10Ϫ9 mbar. Each data point is an averagethese conditions, damage suppression has been demonstrated over the six devices fabricated in one set. All error bars refersuccessfully in Ref. 10. to the reproducibility of the device fabrication at The cryogenic ERDA measurements were performed us- Ͻ10Ϫ9 mbar oxygen. The uncertainty within a set is some-ing a 13.4 MeV Heϩϩ beam produced by our 2–30 MeV what smaller. For an oxygen pressure around 5ϫ10Ϫ6 mbarAVF cyclotron, taking advantage of a broad resonance in the a larger variance even within a single set of samples was4 He͑16O,16O͒4He scattering cross section at 13.4 MeV.11 The observed and will be discussed later. The brightness de-recoil detection angle was set to 30° with respect to the ion creases continuously from an oxygen pressure ofbeam; the angle between the sample normal and the ion Ͻ10Ϫ9 mbar to 10Ϫ7 mbar by about a factor of 2.5. It possi-beam was 70°. To discriminate between scattered He ions bly shows a minimum between 10Ϫ7 and 10Ϫ6 and drops toand recoiled C and O ions, pulse-shape discrimination12 with zero at a pressure of 1ϫ10Ϫ5 mbar. The current reaches alow resistively passivated implanted planar silicon minimum between 10Ϫ7 and 10Ϫ6 mbar and drops also todetectors13 was applied. zero at a pressure of 1ϫ10Ϫ5 mbar. The resulting external efficiency is plotted in Fig. 5. It decreases continuously from Ͻ10Ϫ9 to 5ϫ10Ϫ6 mbar, in total by about a factor of 2. At aIII. RESULTS pressure of 5ϫ10Ϫ6 mbar the current and brightness of the The characteristic of a device without exposure to oxy- devices changes rapidly with the pressure. This will be dis-gen during calcium deposition is shown in Fig. 3. At a volt- cussed below. At this pressure we found two different groupsage of 5 V we achieve a current of 22 mA/cm2 and a bright- in the sets of six samples. Within both groups we foundness of 115 Cd/m2 in the glove box at device temperature of nearly identical results. The samples marked with* have23°. This corresponds to an external efficiency of 0.53 Cd/A. been placed further from the oxygen inlet than the devices Downloaded 08 May 2012 to Redistribution subject to AIP license or copyright; see
  5. 5. J. Appl. Phys., Vol. 90, No. 3, 1 August 2001 Andersson et al. 1379FIG. 5. External efficiency for different oxygen pressures at a bias of 5 V.For the data points at 10Ϫ9 mbar and the measurements marked with* and #holds the same as described in the caption of Fig. 4. FIG. 6. ERDA spectrum ͑black curve͒ of an oxygen-free prepared pLED. The dotted curve represents the RUMP simulation. The energies of recoiled carbon ͑7.54 MeV͒ and ͑6.44 MeV͒ oxygen, present at the surface, aremarked with #. The group marked with # had 4 times less marked with arrows. Also indicated are the features of carbon and oxygen in the OC1C10-PPV film. The aluminum capping is partially oxidized, whichcurrent and 7 times less brightness compared with the other results in the peak at the surface energy of oxygen. The small peak at aboutgroup. Due to their position with respect to the oxygen inlet 6.1 MeV is due to calcium oxide at the interface between calcium andthe samples marked with # experienced a somewhat higher aluminum.oxygen pressure, and we attribute the difference between thegroups thus to the high sensitivity of the device performanceon the oxygen pressure in this range. Differences greater than15% within one set of devices were observed only in thiscase, for all other experiments it was less than that. An ERDA measurement of a device with an oxygenpressure Ͻ10Ϫ9 mbar during calcium deposition is shown inFig. 6. The number of the recoiled particles is plotted againsttheir energy which depends, among others, on the depth inthe target from which they originate. The energy of the re-coiled particle decreases with increasing mass and increasingdepth. The energies of recoiled carbon and oxygen, present atthe surface, are marked in the spectrum. The feature between6.8 and 7.1 MeV is due to carbon from the PPV. The stepbelow 5.8 MeV results from oxygen in the PVV. The peaksat 6.44 and 6.1 MeV are thin oxide layers at the surface ofthe aluminum and at the interface of calcium and aluminum.In a large number of experiments slight variations ͑ϳ30%͒ inthe oxygen areal densities corresponding to the aluminumsurface (ϳ9ϫ1015/cm2 ) and calcium/aluminum interface(ϳ5ϫ1015/cm2 ) have been observed. However, these varia-tions did not significantly influence the I – V-light character-istics of the LEDs. Oxygen within the calcium layer causessignals between 5.8 and 6.1 MeV. The amount of oxygen inthe calcium layer in Fig. 6 is less than 2%, which is thedetection limit in this case. The ERDA spectra for the oxy-gen pressures of 10Ϫ7 and 5ϫ10Ϫ6 mbar are shown in Fig.7. We find an increasing amount of oxygen within the CaOxlayer with increasing oxygen pressure. The values of theamount of oxygen in the CaOx layer determined by simulat- FIG. 7. ERDA spectrum ͑thin black line͒ of a pLED in which the calciuming the ERDA spectra with a modified version of the RUMP cathode was deposited in a background pressure of ͑a͒ 10Ϫ7 mbar and ͑b͒ 5ϫ10Ϫ6 mbar of oxygen. The thick dotted line represents the RUMP simu-code are given in Fig. 8. The simulations are plotted in Figs. lation. The arrows indicate the energies of carbon ͑7.54 MeV͒ and oxygen6 and 7 as dotted lines. The maximum of the oxygen content ͑6.44 MeV͒ present at the surface. Also indicated are the features of carbon͑41 at. %͒ is almost reached at a pressure of 5ϫ10Ϫ6 mbar, in the OC1C10-PPV film and oxygen in the calcium cathode. Downloaded 08 May 2012 to Redistribution subject to AIP license or copyright; see
  6. 6. 1380 J. Appl. Phys., Vol. 90, No. 3, 1 August 2001 Andersson et al. FIG. 10. Current and brightness before and after the first contact with the glove box atmosphere. For the data points at 10Ϫ9 and 5ϫ10Ϫ6 mbar holdsFIG. 8. Oxygen content in the calcium cathode. The amount of oxygen is the same as described in the caption of Fig. 4.determined from the simulations of the ERDA measurements. of 3 and is independent of the oxygen partial pressure duringwhere the LEDs are still operating. The failure of the devices the calcium deposition. Measurements of current and bright-at pressures Ͼ5ϫ10Ϫ6 mbar is due to the fully oxidized and ness at a constant voltage during operation of several hoursnonconducting CaOx layer. Complete oxidation of calcium is within UHV and the glove box are shown in Fig. 11. In UHVreached at a flux ratio onto the surface of oxygen to calcium the LEDs show a gradual decrease in current and brightnessof ϳ3:1, assuming a sticking coefficient of 1 for the calcium in the first 7 h. After that time the performance decreasesin the evaporation process. only slightly within a total operation time up to 60 h. The The onset voltage of the brightness is shown in Fig. 9. LED in the glove box started at a lower current and bright-The onset voltage increases with increasing oxygen pressure ness but shows only a slight decrease. The relative differ-until 5ϫ10Ϫ7 mbar from a value of 2.1–2.3 V, as measured ences in current and brightness between operation in bothfor the samples in the glove box. environments remain constant after operation for 7 h at a All LEDs have been characterized in UHV within 1 h factor of 1.5. Storage of the LEDs in UHV without operationafter their fabrication. The LEDs lose about a factor of 3 in also causes a loss of current and brightness but with a some-current and brightness, when they are transferred from the what smaller factor than the operated LEDs. From Fig. 9 weUHV into the glove box. The time necessary for the transport also see that the onset voltage for the brightness is shifted tois a few minutes. After the first contact with the glove box higher values by about 0.2 V. Further storage of the LEDs inenvironment the LEDs keep the same characteristics after the glove box does not change the performance of the LEDsreintroduction into the UHV chamber as they exhibited in the significantly within 7 days.glove box. The current and brightness due to the differentenvironments at a voltage of 5 V are given in Fig. 10. The IV. DISCUSSIONrelative loss for both current and brightness is about a factor The current, brightness, and efficiency decrease with in- creasing oxygen pressure during deposition of the calcium cathode. We find a minimum in the current, a decrease of theFIG. 9. Onset voltages of the brightness at different oxygen pressures. For FIG. 11. Degradation of the LEDs during the first hours of operation in thethe data points at 10Ϫ9 mbar and the measurements marked with* and # UHV and in the glove box environment. The devices are both fabricated atholds the same as described in the caption of Fig. 4. an oxygen pressure Ͻ10Ϫ9 mbar. Downloaded 08 May 2012 to Redistribution subject to AIP license or copyright; see
  7. 7. J. Appl. Phys., Vol. 90, No. 3, 1 August 2001 Andersson et al. 1381brightness possibly with a small minimum, a continuous loss that of an unoxidized atom which hinders the diffusion ofof the efficiency, and an increase of the onset voltage of the the, calcium into the PPV. This might lead to a sharpening ofbrightness. It can be excluded that an increase of the resis- the CaOx /PPV interface by suppression of the diffusion oftance of the calcium layer with increasing oxygen content individual calcium atoms or ions into the PPV. As a conse-causes the changes of the characteristics. It can be shown quence the trap distribution of holes near the interface asso-that the I – V curves of LEDs with different oxidized calcium ciated with calcium is altered ͑probably lowered͒ which willcannot be shifted onto each other by correcting the applied increase the hole current.voltage for an additional resistor in the CaOx layer. Such a The brightness of the devices decreases more stronglycorrection involves a voltage drop over the partially oxidized than the current with increasing oxygen pressure, which re-calcium layer which is proportional to the current, while the sults in a decreasing efficiency. A reason for the decrease ofadditional resistance of the CaOx layer is kept constant for a the brightness could be a decrease of the area of thecertain LED. PPV/CaOx interface, from which electrons can be injected. If Blom and de Jong developed a model for the charge calcium oxide would grow in islands, the interface of thetransport in organic LEDs for the case that the current is not PPV with CaOx would consist of small areas of fully and oflimited by injection barriers between the electrodes and the unoxidized calcium. The fully oxidized islands will be insu-organic layer.14 They showed that LEDs fabricated with the lating and charges would be injected only from the areassame materials and the same structure as we used have neg- where the calcium is not oxidized. As a result, the active arealigible injection barriers of less than 0.2 eV and that the of the LEDs would be reduced. However, since calcium iscurrent is dominated by hole transport. In their model the very reactive with oxygen, it is very unlikely that the oxygencurrent is described as space charge limited with a tempera- atoms will move over the calcium surface in order to formture and field dependent mobility and a trap distribution for islands of fully oxidized calcium, instead the oxygen will bethe electrons. The I – V curves are fitted by determining nu- chemisorbed immediately.merically the field distribution, the charge carrier concentra- The decrease of the brightness can be explained, in ourtion, and the electron–hole recombination rate in the PPV opinion, by assuming that the electron injection from thelayer. The I – V characteristics with the lowest current— cathode is reduced. The reason could be a reduction of theLEDs fabricated at 5ϫ10Ϫ7 mbar and characterized in the work function of the partially oxidized calcium, which wouldglove box—are nearly identical to that published by Blom also explain an increase of the onset voltage of the bright-and de Jong in Ref. 14, which are space charge limited. All ness. The increase of the work function could be accompa-other LEDs showed a higher current and even a steeper in- nied by the creation of electron traps at the PPV CaOx inter-crease of the current at a bias Ͼ3 V. Only for the set marked face caused by the partially oxidized PPV, which decreaseswith # in Fig. 4 the current was lower. Thus, all the LEDs the current of electrons into the recombination zone of elec-fabricated at oxygen pressures less than or approximately 5 trons and holes and thus decreases the brightness.ϫ10Ϫ6 mbar have a current which is space charge but not The decrease in current and brightness during storage ininjection limited. the UHV and during the transport from the UHV to the glove It seems plausible to explain the decrease of the current box can be explained very well with the change in the tem-with a decrease of the injection of electrons because the cath- perature of the devices and thus the change of the mobility ofode is modified. It can be expected that fewer electrons are the charge carriers using the parameter published by Blominjected from the oxidized calcium, e.g., due to a change of and de Jong.14the work function. However, it should be noted that the de-vices still have I/V curves characteristic for space charge V. CONCLUSIONSlimited charge transport. Since the current shows a pro-nounced minimum at 5ϫ10Ϫ7 the overall I/V behavior can In our setup we are able to fabricate LEDs with a highbe attributed only partially to a blocking of the electron in- reproducibility. We have correlated the oxygen pressure dur-jection with increasing oxygen pressure during deposition of ing the deposition of the calcium cathode with the changes inthe calcium. The reduction of the current simply due to the structure of the LEDs and the performance of the devicesblocking would not allow us to restore the current to the by determining oxygen concentration depth profiles withvalue of the unmodified cathode when the electron injecting cryogenic ERDA of characterized LEDs. The best LEDs areelectrode is nearly oxidized at 5ϫ10Ϫ6 and must be attrib- obtained at an oxygen pressure Ͻ10Ϫ9 mbar, which is inuted to other effects. Since the holes are the major charge contradiction to the results from Salaneck and co-workers.7carriers, a decrease of the current upon oxygen exposure This may be due to different preparation conditions beforemust be also due to either a change in the hole mobility or to depositing the calcium electrode, e.g., by spin coating thea change in the trap distribution of the holes and thus of the PVV in air, which would result in a partially oxidized PPV. Itcharge carrier concentration and the field distribution in the could be also due to differences in the chemical structure ofpolymer layer. A plausible explanation is that the decrease of the used PPV. Comparing our results with those published bythe current for oxygen pressures Ͻ10Ϫ7 mbar is induced by Blom and de Jong show, that the performance of LEDs isoxidation of calcium in the CaOx /PPV interface region. For significantly affected by the residual gas present during theoxygen background pressures near 5ϫ10Ϫ6 mbar the cal- deposition of the calcium electrode in a high vacuum system.cium is nearly oxidized during the deposition process. The We showed that the presence of oxygen during calciummobility of an oxidized calcium atom is probably lower than deposition results in a decrease in brightness and efficiency. Downloaded 08 May 2012 to Redistribution subject to AIP license or copyright; see
  8. 8. 1382 J. Appl. Phys., Vol. 90, No. 3, 1 August 2001 Andersson et al.The degradation process is attributed to the blocking of the ACKNOWLEDGMENTSelectron injection and a decrease in hole mobility close to thePPV/CaOx interface, to an increase in work function of the This work has been sponsored by grants from the Dutchpartially oxidized PPV or the creation of traps. Foundation for Fundamental Research ͑FOM͒ and the Prior- After correction of the I – V-light behavior of the devices ity Program Materials of the Organization for Scientific Re-as measured in the UHV chamber for the difference in tem- search ͑PPM-NWO͒.peratures, we find that device performance after transportinto the glove box is virtually the same as measured in the 1UHV chamber directly after preparation. This shows that the ¨ J. S. Kim, M. Granstrom, R. H. Friend, N. Johansson, W. R. Salaneck, R.devices after preparation are not influenced by the remaining Daik, W. J. Feast, and F. Cacialli, J. Appl. Phys. 84, 6859 ͑1998͒. 2 ¨ F. Steuber, J. Staudigel, M. Stossel, J. Simmerer, and A. Winnacker, Appl.impurities in the glove box, i.e., water and oxygen levels up Phys. Lett. 74, 3558 ͑1999͒.to 1 ppm. Consequently, in our opinion, the device perfor- 3 J. C. Scott, J. H. Kaufman, P. J. Brock, R. DiPietro, J. Salem, and J. A.mance as measured in the glove box is characteristic for the Goitia, J. Appl. Phys. 79, 2745 ͑1996͒.intrinsic properties of the materials used in these devices. 4 S. Karg, M. Meier, and W. Riess, J. Appl. Phys. 82, 1951 ͑1997͒. 5 ¨ ¨ P. Broms, J. Birgersson, N. Johansson, M. Logdlund, and W. R. Salaneck, However, it should be kept in mind that it cannot fully be Synth. Met. 74, 179 ͑1995͒.excluded that during the whole processing and characterizing 6 P. Broms, J. Birgerson, and W. R. Salaneck, Synth. Met. 88, 255 ͑1997͒. ¨ 7 ´of the LEDs in the glove box impurities on a very low level K. Z. Xing, N. Johansson, G. Beamson, D. T. Clark, J.-L. Bredas, and W.can be incorporated in the devices, e.g., during spin coating R. Salaneck, Adv. Mater. 9, 1027 ͑1997͒. 8 ¨ A. Bohler, S. Dirr, H.-H. Johannes, D. Ammermann, and W. Kowalsky,of the polymer layer, and still have a small influence on Synth. Met. 91, 95 ͑1997͒.device performance. Only for devices which can be prepared 9 S. K. So, W. K. Choi, C. H. Cheng, L. M. Leung, and C. F. Kwong, Appl.fully under UHV conditions, e.g., by evaporation of the or- Phys. A: Mater. Sci. Process. 68, 447 ͑1999͒. 10ganic layer in the UHV chamber ͑only feasible for certain M. P. de Jong, L. J. van IJzendoorn, and M. J. A. de Voigt, Nucl. Instrum. Methods Phys. Res. B 161–163, 207 ͑2000͒.oligomers͒, will it be possible to rule out completely the 11 M. K. Mehta, W. E. Hunt, and R. H. Davis, Phys. Rev. 160, 791 ͑1967͒.effect of such impurities. 12 S. S. Klein and H. A. Rijken, Nucl. Instrum. Methods Phys. Res. B 66, For the future we plan to model the I – V-light character- 395 ͑1992͒. 13 Canberra Semiconductor N. V. ͑͒, B-2250 Olen, Bel-istics and to measure hole and electron mobilities in order to gium.determine more precisely the reasons, which cause the de- 14 P. W. Blom and M. J. M. de Jong, IEEE J. Sel. Top. Quantum Electron. 4,scribed effects. 105 ͑1998͒. Downloaded 08 May 2012 to Redistribution subject to AIP license or copyright; see